Your browser is unsupported

We recommend using the latest version of IE11, Edge, Chrome, Firefox or Safari.

ECE.03 – LaWRA: Low-earth orbit Wildfire Response and Analysis CubeSat Mission

Team Members Heading link

  • Victoria Aponte Blizzard
  • Leah Paulding
  • Kayla Vargas

Project Description Heading link

Wildfires are a growing threat to life, infrastructure, and limited public service resources. The Low earth orbit Wildfire Response and Analysis (LaWRA) CubeSat, a 2U computer vision-based cube satellite mission, is proposed to remotely detect, analyze, and alert first responders and wildfire stakeholders to early signs of wildfires in southern California. LaWRA will mitigate wildfire damage and danger by identifying wildfire smoke plumes from low-earth orbit (LEO) and immediately delivering the imagery and location data to fire suppression teams in a cost-effective package. We present the design of the LaWRA payload, a computer vision wildfire and smoke detection system composed of a field-programmable gate array (FPGA) receiving RGB pixel array input from an off-the-shelf camera module. The wildfire and smoke detection are driven by our lightweight implementation of the Swin Transformer, a computationally efficient vision transformer (ViT) ideal for image classification and object detection tasks in wildfire identification. We show that our Swin Transformer-based model computationally outperforms established convolutional neural network methods in wildfire remote sensing. Our method offers reduced training times on large image datasets, making it an ideal choice for real-time satellite imagery applications and we expect our results to serve as a proof-of-technology for the scientific community.